APREX · EST. 2026
A few days in June 2026 served as a wake-up call for the global AI industry. Anthropic, one of America's most prominent AI companies, was ordered by the White House over the weekend to take its newest and most advanced models offline. The reason: Washington demanded that access be blocked for all foreign nationals – including the company's own employees based abroad. Anthropic said it had little choice.
A brutal demonstration of American control
According to The Verge, the action was swift, sweeping, and carried out with very little warning. For users and companies outside the United States, the message was crystal clear: access to the world's most advanced AI can be switched off with the stroke of a pen in Washington.
The incident demonstrated something many already knew but had rarely seen so explicitly: the U.S. not only dominates the development of frontier AI – the American government holds real power over who is permitted to use the technology at all.
The incident served as a warning shot: frontier AI is not neutral infrastructure – it has an address, and that address is Washington D.C.
China moved into the vacuum
For Chinese AI companies, the news was a gift. The market share of Chinese open-source models has already surged – from 1.2 percent of global AI usage at the end of 2024 to nearly 30 percent in December 2025, according to analytics data cited in market reports.
Price is a central factor. Some Chinese models are offered at 43 times lower cost per token compared with equivalent American solutions. Martin Casado, general partner at Andreessen Horowitz, stated in February 2026 that around 80 percent of American startups building on open-source AI are doing so on Chinese models.
DeepSeek is perhaps the most well-known example. The company's R1 model, launched in January 2025, is reported to have been trained for around six million dollars – compared with an estimated 100 million dollars for GPT-4. Moonshot AI's Kimi model is reportedly the engine behind the popular code assistant Cursor. Shares in Zhipu AI rose sharply after the Anthropic restrictions became public.
Chi Zhang, division director at Alibaba Cloud, stated in June 2026 that China – despite a weaker position in compute – is only "100 days behind" the U.S. in frontier AI capabilities.
Europe: Regulation as a competitive advantage
Europe's response to the AI race is differently oriented. France's Mistral AI has positioned itself as an alternative that emphasizes regulatory compliance and human values over raw benchmark results. The UK's Stability AI offers open foundation models for image synthesis with a focus on accessibility.
The EU market for AI was estimated at 66.4 billion dollars in 2024, with projected growth to over 337 billion dollars by 2032. The UK AI market was valued at over 21 billion pounds in 2025.
As a direct consequence of the AI Act, European players have built systems with built-in privacy and ethics guarantees – something that, in a world marked by geopolitical AI uncertainty, may appear as an attractive option for countries that do not want to be at the mercy of American political decisions.
What does this mean for the rest of the world?
A central question arises: how many countries and companies are willing to build critical infrastructure on technology that can be removed by administrative decree? According to Bernstein analysts, Chinese AI labs are well positioned to take a central global role, particularly in consumer, SMB, and emerging markets – precisely because they combine lower prices with geopolitical independence from Washington.
None of the market share figures and claims described here have been independently verified by 24AI, and it is worth noting that AI market growth forecasts have historically been volatile. Nevertheless, the arrows are pointing in a clear direction: the Anthropic weekend may prove to have handed non-American AI players one of their most effective sales pitches ever.
Sources: The Verge, market analytics reports on non-American frontier AI (June 2026)